ORIGINAL ARTICLE

The influence of social relationship on food tolerancein wolves and dogs

Rachel Dale1,2 & Friederike Range1,2 & Laura Stott2 & Kurt Kotrschal2,3 &

Sarah Marshall-Pescini1,2

Received: 30 November 2016 /Revised: 20 June 2017 /Accepted: 21 June 2017 /Published online: 30 June 2017# The Author(s) 2017. This article is an open access publication

AbstractFood sharing is relatively widespread across the animal king-dom, but research into the socio-ecological factors affectingthis activity has predominantly focused on primates. Thesestudies do suggest though that food tolerance is linked to thesocial relationship with potential partners. Therefore, the cur-rent study aimed to assess the social factors which influencefood tolerance in two canids: wolves and dogs. We presentedwolves and dogs with two paradigms: dyadic tolerance testsand group carcass feedings. In the dyadic setting, theaffiliative relationship with a partner was the most importantfactor, with a strong bond promoting more sharing in bothspecies. In the group setting, however, rank was the primaryfactor determining feeding behavior. Although the dominantindividuals of both species defended the carcass more thansubordinates, in the dogs, the subordinates mostly stayedaway from the resource and the most dominant individualmonopolized the food. In the wolves, the subordinates spentas much time as dominant individuals in proximity to, and

feeding from, the carcass. Furthermore, subordinate wolveswere more able to use persistence strategies than the dogswere. Feeding interactions in the wolves, but not dogs, werealso modulated by whether the carcass was on the ground orhanging from a tree. Overall, the social relationship with apartner is important in food distribution in wolves and dogs,but the precise effects are dependent on species and feedingcontext. We consider how the different socio-ecologies of thetwo species may be linked to these findings.

Significance statementDespite the fact that food sharing is relatively widespread inthe animal kingdom, the specific factors underlying whetheran animal will share with a specific individual are little under-stood. When it comes to decisions about food sharing inwolves and dogs, friendship is the deciding factor if it is justtwo of you, but in a bigger group rank position decides youraccess to the spoils. What is more, it seems that rank position-ing is even more important in dogs than wolves as dominantdogs keep the food for themselves while each wolf packmem-ber has a chance to eat. This is the first evidence that theimportance of the social relationship in food sharing is depen-dent on the feeding context in canids.

Keywords Canid . Food tolerance . Social relationship .

Feeding context

Introduction

Food sharing is generally defined as the joint use of amonopolizable food resource by more than one individual(Stevens and Gilby 2004) or the transfer of a food item fromone individual to another (Feistner and McGrew 1989). Foodsharing incurs a cost to the food possessor and, as such, is the

Communicated by C. Soulsbury

Electronic supplementary material The online version of this article(doi:10.1007/s00265-017-2339-8) contains supplementary material,which is available to authorized users.

* Rachel Dalerachel.dale@vetmeduni.ac.at

1 Comparative Cognition, Messerli Research Institute, University ofVeterinary Medicine, Medical University of Vienna, University ofVienna, 1 Veterinaerplatz, 1210 Vienna, Austria

2 Wolf Science Center, Messerli Research Institute, University ofVeterinary Medicine, Vienna, Austria

3 Department of Behavioural Biology, University of Vienna,Vienna, Austria

Behav Ecol Sociobiol (2017) 71: 107DOI 10.1007/s00265-017-2339-8

most commonly seen prosocial behavior in non-human ani-mals, with evidence of its occurrence in a wide variety ofspecies (e.g., Clutton-brock 1991; Vahed 1998; Carter andWilkinson 2015). However, although food sharing has beenobserved in many species, the mechanisms and principles reg-ulating an individual’s choice to share food with another arestill under debate. The primary hypotheses proposed thus farare as follows: (1) kin selection: whereby animals provisionoffspring or those closely related (Feistner and McGrew1989); (2) reciprocity: where an individual relinquishes foodto another in exchange for either past or future benefits fromthe recipient such as receiving food, grooming, sex, or supportin conflict (Brosnan and de Waal 2002); and (3) harassmentavoidance: when a possessor gives up a resource because thisholds a lower cost than defending it or refusing such solicita-tion would incur (Gilby 2006).

In addition to these hypotheses, it has recently been sug-gested that the quality or type of the relationship shared withanother individual may also be an important factor in toler-ance around food sources. Jaeggi and van Schaik (2011) sug-gested that food sharing when under pressure (i.e., when beingsolicited to do so by the potential receiver) is inextricablylinked to the social relationship between the individuals.Rejecting this harassment may have more social costs at cer-tain times or with certain individuals than others; for example,rejecting the harassment of friends or potential mates may leadto loss of future reciprocal benefits. Therefore, they argue thatharassment avoidance can explain the general occurrence offood sharing, but additional explanations are often required toaccount for the specific possessor-recipient combinations seento engage in this activity.

Following this, some authors have proposed that food shar-ing may include a Bsocial assessment^ aspect. Specifically,hypotheses have been proposed to suggest that beggars areoften not primarily interested in food but rather use the re-sponse of the possessor to ascertain information about theirpersonality (the Binformation gathering^ hypothesis; vanNoordwijk and van Schaik 2009) or about the status of theirrelationship with them (the Bassessing-relationships^ hypoth-esis; Goldstone et al. 2016). Related to this, some have sug-gested that food sharing can in fact be used not only to assess,but also to establish or reinforce social bonds (von Bayernet al. 2007; Wittig et al. 2014; Yamamoto 2015). Yamamoto(2015) observed that the reciprocity and harassment avoid-ance hypotheses mainly stem from meat sharing in chimpan-zees, a highly sought after commodity in a competitive spe-cies. Bonobos on the other hand share abundant fruits thatindividuals are able to gain individually, suggesting that rea-sons other than nutritional gain may be at play (see alsoSlocombe and Newton-Fisher 2005).

Although it is still a limited field of research, a growingbody of empirical evidence over the last few years does sup-port this more social view of food sharing, at least in primates.

A strong affiliative bond has been shown to be an importantfactor in food tolerance both in wild and captive populations.For instance, co-feeding patterns significantly correlated withgrooming relationships, but not genetic relatedness, inChacma baboons (King et al. 2011). Although they couldnot rule out an impact of kinship entirely due to the smallnumber of highly related adult dyads, the results do indicatean importance of a strong social bond in co-feeding tolerance.In a captive setting, where more factors can be controlled,including food distribution and knowledge of genetic and so-cial relationships, chimpanzees that were highly affiliatedwitha food possessor were more likely to receive food than thosethat typically avoided the possessor in a group of female chim-panzees (Eppley et al. 2013). Although females shared ahigher proportion of food with some kin than non-kin mem-bers, they only did so when they also shared a close affiliativebond with the related individual. These results were mediatedby perseverance, with closely bonded partners showing moreBbegging^ behaviors than partners with less affiliative bonds,suggesting that in female chimpanzees, persistently requestingfood is only possible if you have a good relationship with thepossessor.

There is some evidence suggesting that the story is not assimple as animals providing more food to Bfriends,^ as insome studies this effect is additionally mediated by rank. Forexample, although grooming predicted food sharing in thestudy by King et al. (2011), dominant baboons were centralto the feeding network, meaning they frequently shared foodpatches with multiple group members, thus giving these indi-viduals more opportunity to demonstrate sharing with partic-ular individuals. Similarly, using a long-term data set,O’Malley et al. (2016) recently found that high-ranking wildfemale chimpanzees spent more time eating meat than low-ranking females. In chimpanzees, meat is acquired by themales and shared with other group members, suggesting thatmales share more meat with high-ranking females. No suchrank difference in female feeding was found in the consump-tion of insects, which can be acquired individually. In addi-tion, although wild Japanese macaques spent more time co-feeding with females which groomed them the most, this re-lationship became even stronger when the analysis was re-stricted to grooming directed from low to high ranking ani-mals (Ventura et al. 2006). This suggests that the low-rankedindividuals may use grooming to Bbuy^ the food tolerance ofdominant animals.

Despite the fact that food sharing occurs in a wide varietyof species, it is clear that the investigations into the factorsaffecting food sharing have predominantly focused on pri-mates. Although some studies have mentioned the potentialinfluence of hierarchy on feeding behavior of non-primates(Hirsch 2007; black bears: Rogers 1987; cats: Bonanni et al.2007; red deer: Appleby 1980; pigs: Held et al. 2010; rooks:Scheid et al. 2008; hyenas: Smith et al. 2007), comprehensive

107 Page 2 of 14 Behav Ecol Sociobiol (2017) 71: 107

assessments of how social relationships influence feeding inother species remain lacking. Therefore, the current studyaimed to assess the conditions which promote, or inhibit, foodsharing in a non-primate taxon, i.e., canids. Canids are partic-ularly interesting because many species, particularly those re-liant on cooperative hunting, share food on a much more reg-ular basis than primates (Moehlman 1989); yet, almost noth-ing is known about how they negotiate the distribution of foodamong group members.

Specifically, gray wolves (Canis lupus) and domestic dogs(Canis lupus familiaris) were our model species since theydiverged relatively recently; yet, their social organizationand feeding ecologies are considerably different. Domesticdogs originally derive from ancestral gray wolves (Lindblad-toh et al. 2005; Frantz et al. 2016), and both species are highlysocial, forming stable social bonds with group members overmultiple years (Harrington and Mech 1982; Bonanni et al.2010). Wolves typically live in monogamous family unitsconsisting of a breeding pair and the offspring from previousyears (Mech and Boitani 2003). They rely on cooperativehunting and breeding (Mech and Boitani 2003; MacNultyet al. 2012), and as such, food sharing is an essential part ofthe survival of pack members, via distribution of large preybrought down together and regurgitation of food by adults forpuppies (Mech et al. 1999). In comparison, free-ranging dogs(whose movements, activities, and reproduction are notconstrained by humans; Cafazzo et al. 2014) live in promis-cuous packs, and although capable of hunting (Manor andSaltz 2004; Silva-Rodríguez and Sieving 2012), they seemto rely more on individual scavenging, predominantly on hu-man refuse, as their primary foraging technique (Butler et al.2004; Vanak and Gompper 2009a).

These differences in foraging techniques may be responsi-ble for recently observed differences in food tolerance be-tween the two species (Marshall-Pescini et al. 2017). In oursimilarly raised and kept sample of captive wolves and dogs,where direct comparisons between wolves and dogs are pos-sible, it has been shown that dogs exhibit more despotic ten-dencies around food, whereby the dominant individual mo-nopolizes the resource, whereas subordinate wolves were ableto challenge their partners for access to the food (Range et al.2015). These findings are in line with long-term wild wolfresearch by Mech (1999), who found that regardless of rank,eachwolf is able to defend their food source. Furthermore, in afree-ranging dog population, Cafazzo et al. (2010) found thataround food, adult high-ranking dogs showedmost aggressionto middle-ranking dogs and relatively little to low-rankingindividuals. Overall, this suggests that rank is a particularlyrelevant factor in mediating food tolerance in dogs, but poten-tially less so in wolves. Furthermore, we have recently shownthat commodity exchange is relevant in discriminate foodsharing, with both species, but particularly dogs, showing afood-for-sex effect, adapting their behavior around food

depending on female reproductive stage (Dale et al. 2017).Hence, it seems that the continued dependence on food shar-ing in wolves and the reduced reliance on this phenomenon indogs may have affected some aspects of how individuals ne-gotiate when and with whom to share.

In the current study, we investigated this aspect further bytaking into account the quality of the social bond betweenindividuals, both in terms of affiliation and rank. Rank andaffiliation quality were determined on the basis of daily obser-vational data of the animals interacting with their pack mem-bers (for ethogram, see supplementary Table S1). The effectsof these factors on food sharing were assessed in two differentcontexts: a controlled dyadic tolerance test where two animalswere simultaneously released onto a food source and a carcasstest where the whole pack was provided with a single carcass.The former test allowed for a measure of each dyad’s behaviorswithout the potential interference of other pack members,whereas the latter allowed for measures of more naturalisticbehaviors over a longer feeding period in a group environment.Additionally, the group tests allowed us to investigate the effectof carcass positioning on feeding behavior as we presented thecarcasses suspended from a tree (hanging) or lying on theground. The aim of the hanging carcass was to increase theeffort required to access the food, simulating a more similarsetting to Bbringing down prey^ in order to assess whether thismay promote more cooperative feeding behavior, or at leastreduce the opportunity for agonistic interactions.

Our primary prediction was that, because social relation-ships are important in both species (Harrington and Mech1982; Bonanni et al. 2010), dyads with a stronger social bondshould show more tolerance around food. In addition, weexpected that in dogs at least, a higher difference in rankwould promote more tolerance around the food than thosedyads close in rank (Cafazzo et al. 2010). Furthermore, basedon previous results (Range et al. 2015), we expected thatdominance would strongly mediate feeding behavior in dogsbut have a reduced or no effect in wolves. Specifically, wepredicted that in dogs, but not wolves, dominant individualswould monopolize the food more, show more aggression, andshow less peaceful sharing than subordinate animals.Furthermore, we predicted that having to pull down the car-cass (i.e., when it was hanging) would increase tolerance anddecrease aggressive interactions in wolves, which heavily relyon group hunting, but have a smaller or no effect on dogs (thatmostly rely on scavenging).

Persistence in solicitation for food affects food sharing be-havior and appears to be mediated by the strength of the socialbond at least in chimpanzees (Eppley et al. 2013).Furthermore, in both wolves and dogs, persistence has beenshown to vary based on female reproductive stage (Dale et al.2017). Therefore, we expected individuals of both species toshow more perseverance to access the food from those withwhom they have a stronger affiliative bond (Eppley et al.

Behav Ecol Sociobiol (2017) 71: 107 Page 3 of 14 107

2013). However, due to the strong influence of rank on foodtolerance in dogs, we also predicted that in this species, but notin wolves, subordinate individuals would show little persis-tence but rather maintain their distance from the foodresource.

Methods

Subjects

Wolf and dog packs at the Wolf Science Center (www.wolfscience.at) were tested. The wolves originate from wildparks in America and Canada. The dogs are mixed breeds andmost originate from shelters in Hungary. The eight dogs of thelatest generation, which were present for the carcass feedingtests (but not the tolerance tests), were bred at the WolfScience Center from Layla/Nia and external, mixed breedmales. All subjects were hand-raised in peer groups from theage of 10 days. They were bottle-fed and later hand-fed byhumans and had continuous access to humans as social part-ners in the first 5 months of their life. After 5 months, theywere introduced into the packs of adult animals and currentlylive in large 2000–8000 m2 enclosures in these groups (seeRange et al. 2015 for more details). As adults, they take part intraining and various behavioral experiments on a daily basis,mostly using commercially available dry food and Bextrawurst^ sausage as rewards. Since divergence, dogs haveevolved different digestive systems and feeding routines thanwolves (Axelsson et al. 2013). In order to account for this andto create similar levels of foodmotivation in both groups, dogsare fed daily using scattered feeding and wolves are fed every2–5 days with individual pieces of meat (e.g., rabbits). Bothspecies had the same amount of experience with feeding testsand monopolizable food sources prior to this study and car-cass feedings are used in public demonstrations on a weeklybasis (one pack per week).

The current experiment tested nine dogs (5F, 4M) and 12wolves (4F, 8M, see Tables 1 and S1 for details). Twomethods

were used to investigate food sharing responses, dyadic foodtolerance tests and carcass feedings in the pack environment.

Food sharing tests

Dyadic tolerance tests

Dyads always comprised two individuals from the same pack.The two animals were released onto a food source at the sametime and tolerance around the food source was measured.Every dyadic combination from each pack completed threetrials.

General set-up

Subjects were placed into separate, but adjacent, side compart-ments where they could see, but not enter, the central enclo-sure (see supplementary movie 1). A sliding door connectedeach compartment to the central enclosure. The animals werealready accustomed to being temporarily separated and tomoving through the doors when opened. The experimenterthen walked into the central enclosure and visibly placed ashallow plastic bowl baited with 10 meat chunks and a hand-ful of dry dog food (20 cm in diameter for dogs and 40 cm forwolves due to their different head sizes) in front of the ani-mals, centrally between them at a distance of 3 m from eachdoor and then left again. The bowl sizes were chosen toreplicate Range et al. (2015) and accounted for the differenthead sizes and body weights of the animals (mean weight:dogs = 24.57 kg, wolves = 40.65 kg, mean head size:dogs = 40.46 cm, wolves = 52.15 cm), such that the bowlswere large enough to allow the animals to eat from the samebowl simultaneously, but were also small enough so that ananimal could easily monopolize it. The meat chunks are ahighly desirable food for both the wolves and dogs and thedry food increased the total volume to allow each trial to lastlonger. The experimenter filmed all trials from the other sideof the fence, at a distance of 5 m from the food location. For all

Table 1 Pack details for the naturalistic tests and number of carcass feedings presented to each

Pack Species # of individuals Individuals Hanging Lying

Kaspar Wolf 3M, 2F Kaspar, Aragorn, Shima, Tala, Chitto 3 3

Geronimo_1 Wolf 3M Geronimo, Amarok, Kenai 1 5

Geronimo_2 Wolf 2M, 1F Geronimo, Wamblee, Yukon 3 3

Meru_1 Dog 4M, 2F Meru, Nia, Gombo, Hiari, Sahibu, Imara 1 1

Meru_2 Dog 2M, 1F Meru, Hiari, Imara 2 1

Nuru_1 Dog 4M, 3F Nuru, Layla, Zuri, Pepeo, Enzi, Panya, Banzai 1 1

Nuru_2 Dog 3M, 3F Nuru, Layla, Zuri, Pepeo, Enzi, Panya 1 2

Asali Dog 2M, 1F Asali, Bora, Banzai 2 4

107 Page 4 of 14 Behav Ecol Sociobiol (2017) 71: 107

trials, subjects were filmed until the food was finished or for amaximum of 2 min.

Individual trials

At the start of each session, each animal was individuallyreleased into the central enclosure through a sliding door andallowed to eat a handful of meat and dry food from the bowlbefore the test began. This was in order to make the animalsaware that food rewards would be placed in the bowl and toensure food motivation. Each subject received one individualtrial before testing and the trial was filmed. If an animal wasnot motivated to eat in the individual trial, the test was notcontinued that day. Although the second individual couldwatch the individual trial, the order of the individual trialsbetween the two subjects was randomized across sessions.Therefore, it was not always the same individual eating first.The handful of food in the individual trial was a small amountrelative to the animal’s daily food intake; as such, it was highlyunlikely that it affected their satiation levels enough to reducemotivation to access a large bowl of meat in the following testtrial.

Test trials

Immediately after the individual trials, the subjects receivedone test trial where, after the experimenter placed the baitedbowl in the central enclosure, the animals were simultaneous-ly released into the enclosure through the sliding doors (seesupplementary movie 1). The test ended when all the foodwasconsumed or after 2 min.

Carcass feedings

For the carcass feedings, the pack was presented with a deercarcass in their home enclosure. The wolves received an entirecarcass and the dogs received a hind leg of a carcass, thedifference in size being due to the differing daily feeding re-quirements of the two species. The animals were removedfrom their home enclosure, a procedure that is normal in theirdaily routine, and the carcass was chained to a tree within theenclosure. The carcass was chained to a tree either lying flaton the ground or suspended off the ground but low enough forthe animals to reach (approximately 50 cm). The person thenleft and the animals were released back into the enclosure(supplementary movie 2). The carcass, and 10 body lengthssurrounding it, was filmed from outside the enclosure throughthe fence at a distance of approximately 5 m, although thisvaried slightly depending on the enclosure. Each session com-menced when the first animal approached within 10 bodylengths of the carcass and ended after exactly 40 min.

Analyses

i. Characterization of social relationships

Regular focal observations of social interactions between packmembers for this study have been conducted at the WolfScience Center since June 2013. Ten-minute long focal animalsamplings (Altmann 1974) were carried out for each individ-ual using the Pocket Observer program (3.2 Software) andwere then imported into the Observer XT 10.5 program (bothfrom Noldus Information Technology, Wageningen,The Netherlands). Each individual’s observations were equal-ly distributed as much as possible, over the time period, aswell as across time of day. During focal sessions, agonistic andaffiliative behaviors were recorded by the Ball occurrencesmethod^ (Altmann 1974). The descriptions of all behavioralpatterns used to determine social relationships can be found inthe ethogram (supplementary materials Table S1). From theseobservations, the hierarchical structure of each pack, as well asthe affiliative and dominance relationship of each dyad withina pack were ascertained (see below). The data is analyzedonce per year as this provides sufficient data to calculate ro-bust social relationship scores, and therefore, the relationshipscores for each test were taken from the corresponding year.

Dominance

Pack hierarchy

The dominance relationships in the packs were analyzedbased on either the submissive or dominance behaviors exhib-ited depending onwhich provided the strongest linearity index(Cafazzo et al. 2010). Based on de Vries’ improved Landau’slinearity index (de Vries 1995), we found that each packshowed a linear hierarchy (i.e., a linearity index of over0.75). The statistical significance of h′ was tested by meansof a two-step randomization test with 10,000 randomizations(de Vries 1995) using MatMan 1.1 (Noldus InformationTechnology, Wageningen, The Netherlands). Pack memberswere therefore ordered using a procedure proposed by deVries for finding a dominance order most consistent with alinear hierarchy (the I&SI (inconsistencies and sum of incon-sistencies) method; de Vries 1998-MatMan program). Thisprovided the values for the ordinal rank measure (with 1 beingthe most dominant) used in the carcass test analyses (seebelow).

Rank distance

In addition to the general hierarchy order of each pack, wewere also interested in the specific dominance relationshipbetween each pair of individuals in a pack. This relationshipcould be compared with the feeding behavior seen in the

Behav Ecol Sociobiol (2017) 71: 107 Page 5 of 14 107

dyadic tolerance tests, when only those two individuals werepresent around the food source. In order to characterize thedyadic dominance relationship, we calculated the David’sscore (Gammell et al. 2003) for each individual and thensubtracted individual A’s score from individual B’s in orderto obtain a value of rank distance for each pair. The advantageof using the David’s score for this measure is that it takes intoaccount the relative strengths of other individuals in the groupwhen calculating the relative dominance score of each animal.

Affiliation

To characterize the quality of the relationship between indi-viduals, an Baffiliation score^ was used (Silk et al. 2013). Thisrepresents the bidirectional frequency of affiliative behaviors(see observation ethogram) exchanged by individuals A andB, divided by observation time (h) for subjects A + B.

Affiliation ¼ Ai þ Bi

Aj þ Bj

where i represents the frequency of affiliative behaviors and jrepresents the total number of observation hours.

ii. Feeding tests

Coding and inter-observer analyses

For both types of test, the videos of each session were codedwith Solomon Coder Beta 15.01.13 (Copyright András Péter,http://solomoncoder.com). The ethograms used for the codingof feeding behavior can be found in the supplementarymaterials (Tables S2 and S3). In the tolerance tests, the behav-ior of both individuals in the dyad was coded, and in thecarcass tests, the behavior of every individual in the packwas coded, as well as interactions with all other partners.

It was not possible to code the data blind because our studyinvolved focal animals requiring individual recognition. Thecoding of the food tolerance tests was carried out by RD with20% coded by LS for reliability (who was blind to the socialrelationships and hypotheses at this stage). Cohen’s kappacoefficient revealed a high level of agreement on all binomialvariables, considering the guidelines by Landis and Koch(1977: >0.75 as excellent and 0.4–0.75 as good; aggression:1.0, peaceful sharing: 0.91, feeding alone: 0.68). The carcassfeedings were coded by LS and RD with 20% coded by bothfor reliability (food monopolization: 0.96, peaceful sharing:0.91, aggression: 0.71, waiting: 0.91, begging: 0.78, arrivefirst: 1, close proximity: 0.77, far proximity: 0.87, defending:0.28, scrounging: 0.39 (after consultation, these two variableshad complete agreement)).

Statistical analyses

To address our main questions, i.e., the influence of affiliativeand dominance relationships on the food-sharing behavior ofwolves and dogs, we carried out a series of models, with theabove factors as explanatory variables and either the occur-rence, frequency, or duration of specific behaviors relating tofood sharing as our dependant variable.

Wolves and dogs could not be directly compared due to animbalance in the number and sex composition of the dyadsand packs (Tables 1 and S1). Nevertheless, the same modelswere run for both species. This method of analysis allowed forthe investigation of the relevant social factors affecting feed-ing behavior in the two canid species.

Tolerance tests

In the tolerance tests, we analyzed whether the duration ofaggression, peaceful sharing (i.e., co-feeding from the samebowl with no signs of threat), and food monopolization (i.e.,one individual feeding alone from the bowl; seesupplementary materials for detailed definitions) were affect-ed by the rank distance and affiliative relationship between theindividuals in a dyad. Accordingly, we ran generalized linearmixed models (glmer function in the lme4 package) with aGaussian distribution and logit link function. Our dependantvariables were the durations of the behaviors/behavioral cate-gories described above (divided by trial length), and the ex-planatory variables were the affiliation score and rank distanceas measures of the quality of the relationship of each dyad. Assome subjects were tested in multiple dyads and each dyadwas tested repeatedly, we created a variable whereby eachindividual-dyad combination was given a unique identifier,and this was inserted as the random effect in the models.

In addition, due to a higher variation in the likelihood ofpeaceful co-feeding occurring at all in the wolves (see results),we also ran a GLMM with a binomial distribution to assesswhether the likelihood of occurrence of peaceful co-feedingwas affected by the social relationship. The same fixed andrandom effects were included in the model.

For all GLMMs, model assumptions were met. The follow-ing construct depicts the model used for all GLMMs for thetolerance tests:

Responseijk∼affiliationi þ rank distance j þ animalij þ eijk

In this model, affiliationi is the fixed effect of a dyad’saffiliation score and rank_distancej is the fixed effect of therank distance between two individuals of a dyad. Animalij isthe random animal within dyad effect with mean zero and eijkis the random residual with mean zero.

107 Page 6 of 14 Behav Ecol Sociobiol (2017) 71: 107

Carcass tests

As with the dyadic tests, dogs and wolves were analyzedseparately. In the carcass tests, two types of behaviors werecoded: (1) behaviors that had a recipient (i.e., peaceful shar-ing, aggression, waiting, begging, and defending) and (2) be-haviors that did not have a recipient (food monopolization,scrounging, arriving first at the carcass, proximity to the car-cass) (see supplementary ethogram for detailed descriptions ofeach behavior). All behaviors that occurred within 10 bodylengths of the carcass were coded.

As in the tolerance tests, for behaviors that had a recipient,the fixed effects for all models were affiliation score and or-dinal rank. Carcass position (hanging or lying) was also in-cluded as a fixed effect for the peaceful sharing and aggressionmodels. Ordinal rank was used rather than rank distance be-tween two individuals (as for the dyadic testing) since ourmain question in this group context was whether the behaviorsexhibited during feeding were affected by the individuals’position in the pack hierarchy. A variable whereby eachindividual-dyad-pack combination was given a different namewas included as the random effect.

For the behaviors that did not have a recipient, we assessedthe impact of hierarchical position in the pack on theoccurrence/duration of these variables. For these variables, arandom factor with each individual-pack combination wascreated and ordinal rank was the fixed effect.

For all behaviors, if its occurrence was relatively frequent(i.e., occurred in more than 50% of data points) and met themodel assumptions, either the frequency (aggression, carcass

proximity) or the duration (peaceful feeding, food monopoli-zation) was entered in the model as the dependant variable,whereas if the behavior was infrequent, analyses were carriedout on the likelihood of its occurrence (waiting, begging,defending, scrounging, arriving first). Accordingly, GLMMswith a Gaussian distribution (package: lmer) were run forduration variables, glmmPQL for frequencies (package:nlme), and GLMM with a binomial distribution for 1/0 vari-ables (package: lmer).

All analyses were carried out in R version 3.2.2 (R CoreTeam 2015).

Results

Tolerance tests

In both wolves and dogs, peaceful sharing was mediated bythe social relationship between individuals. In dogs, peacefulco-feeding occurred on 97% of trials, but how long it lastedfor was strongly affected by the relationship with the partner;the higher the affiliation score, the more time dyads spentpeacefully sharing (Fig. 1), but no effect of rank distanceemerged. In wolves, peaceful co-feeding occurred on 83%of trials, and here, the duration was not affected by affiliationor rank distance. However, due to this higher variation inwhether or not sharing occurred, we also analyzed the likeli-hood of peaceful sharing occurring in a trial and this washigher in dyads with a higher affiliation score and a higherrank distance. Full model results are presented in Table 2.

Fig. 1 The higher the affiliationscore of a dog dyad, the longerthey peacefully co-fed for in thetolerance tests

Behav Ecol Sociobiol (2017) 71: 107 Page 7 of 14 107

Neither the affiliation score nor the rank distance of a dyadaffected the durations of aggression or food monopolizationshown in either species (Table 2).

Carcass tests

The typical pattern of a carcass feeding was generally the sameacross sessions and packs but varied according to the speciestested (see supplementary movie 2 and Fig. 2 for examples ofhow the feedings looked in each species). The full model outputsfor the carcass tests are presented in Tables 3, 4, and 5.

More dominant wolves were more likely than lower rank-ing wolves to defend the carcass, preventing others from ap-proaching despite not eating themselves, and were more likelyto arrive first at the carcass at the start of the session, but werenot more likely to monopolize the carcass for feeding thanlower ranking animals. Moreover, rank did not affect howmuch time wolves spent close (<1 body length) or far (>10body lengths) from the carcass (Fig. 2).

On the contrary, the dominant dogs, although not morelikely to arrive first, spent significantly more time monopoliz-ing the carcass (specifically the most dominant dog; Fig. 3)and were more likely to defend it from others than subordinate

dogs. In fact, in 40 min of carcass test, subordinate dogs (i.e.,all but the most dominant individual) spent an average of lessthan 50 s feeding alone at the carcass (almost 10 times lessthan subordinate wolves, mean = 462 s). Furthermore, prox-imity to the carcass was significantly affected by rank in thedogs. More dominant dogs were significantly more likely tospend time close to the carcass than subdominants, whereassubordinate dogs spent more time at a great distance from theresource (>10 body lengths; Fig. 2). In fact, as seen in Fig. 2,this was not a gradual effect of rank, with distance increasingas rank position decreased, but rather that the most dominantindividual monopolized the proximate area around the carcassand all other individuals mostly stayed away.

The duration of peaceful sharing in wolves was not affectedby affiliation score, ordinal rank position, or carcass position(Table 3). However, aggression was affected by ordinal rankand carcass position. Higher ranked individuals showed moreaggression than subordinates, but interestingly, aggression wasless frequent when the carcass was hanging than when it waslying. Affiliation score, however, did not affect the amount ofaggression shown towards a particular partner (Table 3).

In the dogs, the duration of peaceful sharing and frequencyof aggression were strongly affected by rank position, with

Table 2 Full model results from the tolerance tests for each variable and species

Wolves Dogs

Affiliation Rank distance Affiliation Rank distance

Duration Duration

Peaceful sharing χ2 = 0.75 (1), p = 0.39 χ2 = 1.25 (1), p = 0.26 χ2 = 7.14 (1), p = 0.008 χ2 = 2.05 (1), p = 0.15

Likelihood

Peaceful sharing χ2 = 6.99 (1), p = 0.008 χ2 = 11.57 (1), p < 0.001

Duration Duration

Food monopolization χ2 = 1.43 (1), p = 0.2 χ2 = 2.46 (1), p = 0.12 χ2 = 1.72 (1), p = 0.19 χ2 = 0.04 (1), p = 0.95

Duration Duration

Aggression χ2 = 0.22 (1), p = 0.64 χ2 = 0.31 (1), p = 0.58 χ2 = 1.35 (1), p = 0.25 χ2 = 0.97 (1), p = 0.32

Fig. 2 A typical picture of how acarcass feeding session looks indogs (left) and wolves (right).Reddots represent the most dominantindividual and blue dots all otherpack members. Each dot is whereone individual spent most of theirtime (the greatest number ofproximity scans during thesession) in one session. The ringsdenote distance from the carcass:0–1, 1–5, and 5–10 body lengths

107 Page 8 of 14 Behav Ecol Sociobiol (2017) 71: 107

higher ranking dogs spending less time peacefully co-feedingthan lower ranked individuals and mainly the most dominantindividual showing aggression. Neither the duration of peace-ful sharing nor the frequency of aggression were affected bycarcass position (hanging vs lying) or affiliation score in thedogs (Table 3).

Begging and waiting (coded as measures of persistence)were not affected by ordinal rank or affiliation score, norwas the likelihood of scrounging affected by rank position ineither species (Table 4).

From the results above, it appears that in the dogs, thesubordinate animals were staying away from the carcass alto-gether (Fig. 2). However, the wolves did not show such aneffect of rank on proximity to the carcass. Therefore, we wereinterested in what it was that the subordinate wolves weredoing that the dogs were not. For only the subordinate animals

(here considered as every individual other than the most dom-inant pack member), we compared the wolves and dogs inscrounging, begging, and waiting. There was no effect of spe-cies on the likelihood of begging but subordinate wolves weresignificantly more likely to scrounge and wait than subordi-nate dogs (Table 5).

Discussion

In sum, results from the dyadic tolerance tests indicate that inboth species, the affiliative relationship with a partner is themost relevant factor in dictating the level of tolerance shownaround the food source, with more peaceful sharing occurringin dyads with higher affiliation scores. Unsurprisingly, thepicture is a little more complex in the carcass tests, where

Table 3 Results of the duration of peaceful sharing, food monopolization, and the frequency of aggression in the carcass feedings dependent onaffiliation score (continuous score based on observations), rank (linear position), or carcass position (hanging vs lying)

Wolves Dogs

Variable Fixed effect χ2 df p value χ2 df p value

Peaceful sharing(duration)

Affiliation score 2.27 1 0.13 0.03 1 0.64

Ordinal rank 0.0.8 1 0.77 17.08 1 0.0001

Carcass position 0.03 1 0.86 2.29 1 0.13

Wolves Dogs

Fixed effect χ2 df p value χ2 df p value

Aggression(frequency)

Affiliation score 0.55 1 0.46 2.00 1 0.16

Ordinal rank 7.00 1 0.008 30.77 1 0.0001

Carcass position 6.12 1 0.01 0.25 1 0.61

Wolves Dogs

Fixed effect χ2 df p value χ2 df p value

Food monopolization Ordinal rank 0.36 1 0.55 5.45 1 0.02

Table 4 Model outputs for the effects of ordinal rank (and affiliation score) on the following variables

Variable Fixed effect Wolves Dogs

Frequency χ2 df p value χ2 df p value

#scans spent <1 body length from carcass Ordinal rank 0.46 1 0.50 6.73 1 0.009

#scans spent >10 body lengths from carcass Ordinal rank 2.41 1 0.12 5.33 1 0.02

Binomial χ2 df p value χ2 df p value

Waiting Ordinal rank 1.48 1 0.22 0.07 1 0.79

Affiliation score 0.44 1 0.51 0.1 1 0.76

Begging Ordinal rank 0.04 1 0.84 0.15 1 0.70

Affiliation score 0.92 1 0.34 0.69 1 0.41

Defending Ordinal rank 3.64 1 0.05 4.60 1 0.03

Affiliation score 0.15 1 0.69 1.27 1 0.26

Scrounging Ordinal rank 0.18 1 0.67 0.03 1 0.86

Arrive first Ordinal rank 4.54 1 0.03 2.52 1 0.11

Behav Ecol Sociobiol (2017) 71: 107 Page 9 of 14 107

the whole pack was present. However, the primary differencewith the dyadic tolerance tests was that affiliation did notemerge as predictive of food tolerance in this context. Rank,on the other hand, was the more important factor in determin-ing an individual’s behavior around the food in the groupsetting. For wolves, but not for dogs, the hanging carcass alsoaffected their behavior by reducing the frequency of aggres-sive interactions.

The results from the dyadic tolerance tests support our pri-mary prediction, demonstrating that when just two individualsare presented with a situation of a small food source, thestronger their affiliative bond the more they engaged in peace-ful sharing. It is logical that animals should choose to co-feedwith partners who are most likely to show tolerance, than aless close affiliate, who may challenge you or incite conflict(de Waal 2000; Heesen et al. 2014). Interestingly though, howthis effect of affiliation emerged appears to differ between thetwo species. In the dogs, co-feeding occurred on 97% of trials,but the closeness of the social bond affected the time spent co-feeding. In the wolves, no such effect was seen on the durationof peaceful sharing, but it seems that their decision aboutwhether or not to share at all was based on the affiliativerelationship, as those with a higher affiliation score were morelikely to peacefully share. At this stage, we cannot concludewhether this difference is due to a species difference in howthe animals decide whether/how to share or whether this isdriven by the imbalance in sex compositions between the spe-cies, as we had mostly male-female pairs in the dogs but allpossible sex compositions in the wolves.

In the wolves, we also found that those dyads with a higherrank distance between the individuals were more likely topeacefully share. This may be due to age, with young individ-uals often occupying low dominance ranks and/or it may bebecause those lower in rank present less competition to thestatus position/mating opportunities of high-ranked individ-uals. These findings are in line with those by Cafazzo et al.(2010) in free-ranging dogs that high ranking dogs oftenshowed aggression to mid-ranking individuals but displayeda certain level of tolerance towards low-ranking individuals.However, interestingly, the results are in contrast to findings inprimates, where those close in rank are predicted to feed to-gether (Seyfarth 1977; Matsumura and Okamoto 1997; Tiddiet al. 2012), even after controlling for kinship (de Waal 1991;but see Kapsalis and Berman 1996). The fact that we did notfind this effect in the dogsmay reflect a species effect and/or ispossibly because, differently from wolves, dog pairs weremostly male-female and factors other than rank distance mayalso play a role in their feeding behavior (Dale et al. 2017).

The affiliative relationship shared by individuals did notaffect any of the coded behaviors around the food in the car-cass context. This may be because in a group setting, youcannot necessarily co-feed with your Bfriends^ as they maynot be permitted by other pack members to access the re-source. In this situation, it was the rank that predominantlyaffected the behavior of the animals. In the wolves, the indi-viduals higher in the hierarchy showed more aggression, weremore likely to arrive first at the carcass, and were more likelyto defend it (i.e., not feeding, but preventing others from ap-proaching) than pack members lower in rank. However, theywere not more likely to spend time close to the carcass thanmore subordinate individuals nor did they monopolize theresource more than others, suggesting that the dominantwolves were trying to control access to the food, but not mo-nopolizing it for themselves (Noë et al. 1980). Indeed, it hasbeen found that in rooks, food offering acted as a costly signalof rank, whereas tolerated co-feeding was explained by reci-procity and formation of social bonds (Scheid et al. 2008).

Table 5 Results from the glmms comparing subordinate wolves anddogs on the likelihood of the occurrence of the following variables

Binomial Fixed effect χ2 df p value

Scrounging Species 5.85 1 0.01

Waiting Species 3.68 1 0.05

Begging Species 0.001 1 0.97

Fig. 3 Mean duration of foodmonopolization (sec) by the mostdominant member vs all otherpack members. * < 0.05. Errorbars represent the standard errorof the mean

107 Page 10 of 14 Behav Ecol Sociobiol (2017) 71: 107

These findings corroborate with our results in wolves thataffiliation was important in peaceful sharing but rank deter-mined defense of the carcass. Furthermore, this interpretationcorresponds with data from wild wolves which suggests thatthe dominant individuals choose whom to allot food to, andthis in turn ensures the survival of the whole pack (Mech1999).

Although rank was also an important factor in dog carcassfeeding behavior, it was in a different manner to that of thewolves. Dominant dogs were, like the wolves, more likely todefend the carcass and showed more aggression than subordi-nate individuals. However, this appeared to be in order tomaintain and feed on the carcass themselves, since they werealso less likely to co-feed and muchmore likely to monopolizeand spend time in close proximity to the resource than moresubordinate individuals, despite being no more likely thansubordinates to arrive first. These results support our secondhypothesis that feeding behavior in dogs would be more de-pendent on rank than in wolves.

Furthermore, the results also corroborate those ofRange et al. (2015), who found that in a dyadic context,feeding was mediated by rank in the dogs, but not in thewolves. Our results from the tolerance tests of the currentstudy are not directly comparable with those of the previ-ous research as we had mostly male-female dog pairsduring this phase of the research. However, results fromthe carcass tests (where same sex relationships were in-cluded) support the interpretation that dogs show muchless tolerance, and a steeper hierarchy around foodsources than wolves with the most dominant dog mostlymonopolizing the resource and the rest of the pack spend-ing most of the session more than 10 body lengths awayfrom the carcass. So, even though the results refer to theanimals’ ordinal rank position in the pack, a closer look atthe raw data (highlighted in Figs. 2 and 3) shows that it isonly the most dominant dog that monopolizes the foodsource. In contrast, the wolves did not show an effect ofrank on proximity to the carcass during a session, mean-ing that both dominant and subordinate animals were aslikely to be close to it. Dubuc and Chapais (2007) suggestthat an individual’s spatial position during group feedingaffects potential feeding gains, but the position of subor-dinates may be mediated by the tolerance of the domi-nants (see also King et al. 2011). In line with this, ourresults suggest that subordinate wolves, but not dogs, aretolerated close to the carcass and are more likely to gainaccess to it.

It is clear that subordinate dogs (here considered all but themost dominant individual) tended to avoid the carcass alto-gether, but in order to ascertain what it was that subordinatewolves were doing, we compared the subordinate individualsof both species. We found that wolves were more likely toshow persistence behaviors (namely waiting and scrounging)

than the dogs were. Eppley et al. (2013) suggested that per-sistence is not a strategy available to all individuals; toleranceis required by a food possessor before an individual will dem-onstrate persistence behaviors. Therefore, subjects tend toBbeg^ from possessors with whom they have a strong socialbond. We had predicted that this would also be the case in thecurrent study, but here, the restricted use of persistence seemsto be at play at the species level instead, with wolves beingmore able to make use of persistence as a strategy to access thefood than dogs. Subordinate wolves also had as much successas dominant wolves in monopolizing the carcass, further sug-gesting that rank is less strictly enforced in the feeding contextin wolves. These findings support our predictions of higherlevels of persistence from subordinate wolves and the avoid-ance of the resource by the subordinate dogs.

Another strategy suggested as a tactic for subordinates togain access to food is arriving first at the resource (Dubuc andChapais 2007). Dubuc and Chapais (2007) found that in long-tailed macaques, rank did not determine the order of arrival ata feeding site, but early arrival did allow for more food con-sumption. Interestingly, our results appear to contradict this aswe found that dominant wolves were more likely to arrivefirst, but did not monopolize the food more than subordinates.In contrast, subordinate dogs were as likely to arrive first at thecarcass but nevertheless were unable to monopolize the food.This suggests a potential difference in the use of first arrival,with our animals appearing not to use this as a tactic for re-source access.

Another factor that we considered was how the position ofthe carcass may affect the tolerance of these two species.Partially supporting our final prediction, wolves (but not dogs)showed less aggression when the carcass was suspended froma tree than when it was lying on the ground, but the positiondid not affect the duration of peaceful sharing. This potentiallysuggests that a hanging carcass promotes cooperation ratherthan competition, as would be required to bring down prey inwild situations and in turn, the process of bringing down preymay be a factor that helps maintain low aggression levels inwild wolves. Alternatively, when the carcass is hanging, thewolves may be so distracted by the desire to pull it down, thatthey have less focus on the other individuals around.Regardless of the underlying motive, this finding potentiallyaffects the management of captive wolf populations, suggest-ing that feedings could include this feature as it may promotemore tolerance. From an evolutionary perspective, the factthat it did not affect the dogs may be because, although theycooperate with humans, they are no longer as cooperative withconspecifics (Range and Viranyi 2015; Marshall-Pescini et al.2017).

It could be argued that the different feeding routines andbowl and carcass sizes used with the wolves and dogs affectedthe results. However, the feeding routines are different in orderto match the requirements of each group. Wolves and dogs

Behav Ecol Sociobiol (2017) 71: 107 Page 11 of 14 107

show differences in feeding ecologies and digestive systems;wolves have a meat-based diet and typically eat every fewdays since they are required to hunt for their meals, whereasdogs scavenge on a daily basis and eat a more starch-rich diet(Vanak and Gompper 2009b; Axelsson et al. 2013).Additionally, the bowl and carcass sizes were chosen accord-ing to the body sizes of the animals, such that in both species,the resource could be shared, but was small enough to bemonopolized if an individual so desired. Furthermore, in nei-ther species does the daily feeding involve restricted,monopolizable resources, and both groups had the sameamount of experience with feeding tests prior to this study.However, although we feel that these factors do not preventa comparison between wolves and dogs, we are aware thatthese factors are different from the feeding situations of free-ranging individuals. It is interesting to note that despite thelimitations of testing and generalizing from captive popula-tions, the results from both species do align with their respec-tive social organizations and feeding ecologies in free-rangingsettings (Marshall-Pescini et al. 2017). Because wolves arecooperative hunters (MacNulty et al. 2012), it is essential forthe survival of the pack that every member is able to accessfood resources, regardless of their rank. Dogs, on the otherhand, rely predominantly on solitary scavenging from humanwaste (Butler et al. 2004; Vanak and Gompper 2009a); there-fore, it is not necessary to allow other group members to ac-cess a resource, and in fact, this may even harm your ownfitness. These differing feeding ecologies appear to bereflected in our results from the carcass tests, whereby subor-dinate wolves are able to exhibit more persistence and feedfrom the carcass considerably more than subordinate dogs.

Overall, what emerges from these tests is that the socialrelationship with a partner affects an individual’s behavioraround a food source in both wolves and dogs. However, thiseffect is mediated by context, with the affiliative relationshipbeing the driving predictor of tolerance in a restricted, dyadicsetting but rank overiding this in the group tests, where thewhole pack is present. Furthermore, the wolves’ reliance oncooperation allows all pack members access to the food,whereas dogs showmore despotic behavior around food, withthe dominant individual monopolizing the carcass. These testsreveal that social factors are also important in determining thefeeding behavior of canids and highlight the benefits of usingmultiple contexts and species in order to ascertain the socio-ecological factors driving the distribution of food in socialgroups.

Acknowledgments Open access funding provided by University ofVeterinary Medicine Vienna. The Wolf Science Center was establishedbyKurt Kotrschal, Friederike Range, and Zsófia Virányi, andwe thank allthe helpers who made this possible hence indirectly supporting this re-search. Additionally, we thank all the staff and students of the WolfScience Center for their help with the tests and care of the animals. Wealso thank Stephan Reber for his statistical advice. Thanks to two

anonymous reviewers for their thoughtful comments on an earlier versionof the manuscript. This work was supported by the European ResearchCouncil under the European Union’s Seventh Framework Programme(FP/2007–2013/ERC Grant Agreement no. 311870). We further thankmany private sponsors, including Royal Canin, for financial supportand the Game Park Ernstbrunn for hosting the WSC.

Compliance with ethical standards

Funding This work was supported by the European Research Councilunder the European Union’s Seventh Framework Programme (FP/2007–2013/ERC Grant Agreement no. 311870).

Conflict of interest The authors declare that they have no conflict ofinterest.

Animal welfare and ethical approval All applicable international,national, and/or institutional guidelines for the care and use of animalswere followed. The research was discussed and approved by the institu-tional ethics committee at the University of Veterinary Medicine, Vienna,in accordance with GSP guidelines and national legislation (03/01/97/2014).

Informed consent Informed consent was not required as no humanparticipants were involved.

Data availability All data generated or analyzed during this study areincluded in this published article [and its supplementary informationfiles].

Open Access This article is distributed under the terms of the CreativeCommons At t r ibut ion 4 .0 In te rna t ional License (h t tp : / /creativecommons.org/licenses/by/4.0/), which permits unrestricted use,distribution, and reproduction in any medium, provided you giveappropriate credit to the original author(s) and the source, provide a linkto the Creative Commons license, and indicate if changes were made.

References

Altmann J (1974) Observational study of behavior: sampling methods.Behaviour 49:227–267

Appleby MC (1980) Social rank and food access in red deer stags.Behaviour 74:294–309

Axelsson E, Ratnakumar A, Arendt ML, Maqbool K, Webster MT,Linblad-Toh K (2013) The genomic signature of dog domesticationreveals adaptation to a starch-rich diet. Nature 495:360–364

Bonanni R, Cafazzo S, Fantini C, Pontier D, Natoli E (2007) Feedingorder in an urban feral domestic cat colony: relationship to domi-nance rank, sex and age. Anim Behav 74:1369–1379

Bonanni R, Cafazzo S, Valsecchi P, Natoli E (2010) Effect of affiliativeand agonistic relationships on leadership behaviour in free-rangingdogs. Anim Behav 79:981–991

Brosnan SF, de Waal FBM (2002) A proximate perspective on reciprocalaltruism. Hum Nat 13:129–152

Butler JRA, Du Toit JT, Bingham J (2004) Free-ranging domestic dogs(Canis familiaris) as predators and prey in rural Zimbabwe: threatsof competition and disease to large wild carnivores. Biol Conserv115:369–378

Cafazzo S, Bonanni R, Valsecchi P, Natoli E (2014) Social variablesaffecting mate preferences, copulation and reproductive outcomein a pack of free-ranging dogs. PLoS One 9:e98594

107 Page 12 of 14 Behav Ecol Sociobiol (2017) 71: 107

Cafazzo S, Valsecchi P, Bonanni R, Natoli E (2010) Dominance in rela-tion to age, sex, and competitive contexts in a group of free-rangingdomestic dogs. Behav Ecol 21:443–455

Carter GG, Wilkinson GS (2015) Social benefits of non-kin food sharingby female vampire bats. Proc R Soc B 282:20152524

Clutton-brock TH (1991) The evolution of parental care. PrincetonUniversity Press, Princeton

Dale R, Marshall-Pescini S, Range F (2017) Do females use their sexualstatus to gain resource access? Investigating food-for-sex in wolvesand dogs. Curr Zool 63:323–330

de Vries H (1995) An improved test of linearity in dominance hierarchiescontaining unknown or tied relationships. Anim Behav 50:1375–1389

de Vries H (1998) Finding a dominance order most consitent with a linearhierarchy: a new procedure and review. Anim Behav 55:827–843

deWaal FBM (1991) Rank distance as a central feature of rhesus monkeysocial organization: a sociometric analysis. Anim Behav 41:383–395

de Waal FBM (2000) Primates-a natural heritage of conflict resolution.Science 289:586–590

Dubuc C, Chapais B (2007) Feeding competition inMacaca fascicularis:an assessment of the early arrival tactic. Int J Primatol 28:357–367

Eppley TM, Suchak M, Crick J, de Waal FBM (2013) Perseverance andfood sharing among closely affiliated female chimpanzees. Primates54:319–324

Feistner A, McGrew W (1989) Food sharing primates: a critical review.In: Seth PK, Seth S (eds) Perspectives in primate biology, vol 3.Today & Tomorrow’s Printers and Publishesrs, New Delhi, pp 21–36

Frantz LAF, Mullin VE, Pionnier-Capitan M et al (2016) Genomic andarchaeological evidence suggests a dual origin of domestic dogs.Science 352:1228–1231

Gammell MP, de Vries H, Jennings DJ, Carlin CM, Hayden TJ (2003)David’s score: a more appropriate dominance ranking method thanClutton-Brock et al.’s index. Anim Behav 66:601–605

Gilby IC (2006) Meat sharing among the Gombe chimpanzees: harass-ment and reciprocal exchange. Anim Behav 71:953–963

Goldstone LG, Sommer V, Nurmi N, Stephens C (2016) Food beggingand sharing in wild bonobos ( Pan paniscus): assessing relationshipquality. Primates 57:367–376

Harrington FH, Mech LD (1982) Patterns of homesites attendance in twoMinnesota wolf packs. In: Harrington FH, Paquet PC (eds) Wolvesof the world: perspectives of behaviour, ecology and conservation.Noyes Publications, Park Ridge, pp 81–105

Heesen M, Rogahn S, Macdonald S, Ostner J, Schülke O (2014)Predictors of food-related aggression in wild Assamese macaquesand the role of conflict avoidance. Behav Ecol Sociobiol 68:1829–1841

Held SD, Byrne RW, Jones S, Murphy E, Friel M, Mendl MT (2010)Domestic pigs, Sus scrofa, adjust their foraging behaviour to whomthey are foraging with. Anim Behav 79:857–862

Hirsch BT (2007) Costs and benefits of within-group spatial position: afeeding competition model. Q Rev Biol 82:9–27

Jaeggi AV, van Schaik CP (2011) The evolution of food sharing in pri-mates. Behav Ecol Sociobiol 65:2125–2140

Kapsalis E, Berman CM (1996)Models of affiliative relationships amongfree-ranging rhesus monkeys (Macaca mulatta) I. Criteria KinshipBehav 133:1209–1234

King AJ, Clark FE, Cowlishaw G (2011) The dining etiquette of desertbaboons: the roles of social bonds, kinship, and dominance in co-feeding networks. Am J Primatol 73:768–774

Landis J, Koch G (1977) The measurement of observer agreement forcategorical data. Biometrics 33:159–174

Lindblad-Toh K, Wade CM, Mikkelsen TS, Karlsson EK, Jaffe DB,Kamal M, Mauceli E (2005) Genome sequence, comparative

analysis and haplotype structure of the domestic dog. Nature 438:803–819

MacNulty DR, Smith DW, Mech LD, Vucetich JA, Packer C (2012)Nonlinear effects of group size on the success of wolves huntingelk. Behav Ecol 23:75–82

Manor R, Saltz D (2004) The impact of free-roaming dogs on gazelle kid/female ratio in a fragmented area. Biol Conserv 119:231–236

Marshall-Pescini S, Cafazzo S, Virányi Z, Range F (2017) Integratingsocial ecology in explanations of wolf-dog behavioral differences.Curr Opin Behav Sci 16:80–86

Matsumura S, Okamoto K (1997) Factors affecting proximity amongmembers of a wild group of moor macaques during feeding, mov-ing, and resting. Int J Primatol 18:929–940

Mech LD (1999) Alpha status, dominance, and division of labor in wolfpacks. Can J Zool 77:1196–1203

Mech LD, Boitani L (2003)Wolves: behavior, ecology, and conservation.University of Chicago Press, Chicago

Mech LD, Wolf PC, Packard JM (1999) Regurgitative food transferamong wild wolves. Can J Zool 77:1192–1195

Moehlman PD (1989) Intraspecific variation in canid social systems. In:Gittleman JL (ed) Carnivore behavior, ecology, and evolution.Springer, Boston, pp 143–163

Noë R, deWaal FBM, van Hooff JARAM (1980) Types of dominance ina chimpanzee colony. Folia Primatol 34:90–110

O’Malley RC, Stanton MA, Gilby IC, Lonsdorf EV, Pusey A, MarkhamAC, Murray CM (2016) Reproductive state and rank influence pat-terns of meat consumption in wild female chimpanzees (Pan trog-lodytes schweinfurthii). J Hum Evol 90:16–28

Core Team R (2015) R: a language and environment for statistical com-puting. R Foundation for Statistical Computing, Vienna, Austriahttp://www.Rproject.org

Range F, Ritter C, Virányi Z (2015) Testing the myth: tolerant dogs andaggressive wolves. Proc R Soc B 282:20150220

Range F, Virányi Z (2015) Tracking the evolutionary origins of dog-human cooperation: Bthe canine cooperation hypothesis^. FrontPsychol 5:1582

Rogers LL (1987) Effects of food supply and kinship on social behaviour,movements, and population growth of black bears in northeasternMinnesota. Wildlife Monogr 97:1–72

Scheid C, Schmidt J, Noë R (2008) Distinct patterns of food offering andco-feeding in rooks. Anim Behav 76:1701–1707

Seyfarth RM (1977) A model of social grooming among adult femalemonkeys. J Theor Biol 65:671–698

Silk J, Cheney D, Seyfarth R (2013) A practical guide to the study ofsocial relationships. Evol Anthropol 22:213–225

Silva-Rodríguez EA, Sieving KE (2012) Domestic dogs shape thelandscape-scale distribution of a threatened forest ungulate. BiolConserv 150:103–110

Slocombe KE, Newton-Fisher NE (2005) Fruit sharing between wildadult chimpanzees (Pan troglodytes schweinfurthii): a socially sig-nificant event? Am J Primatol 65:385–391

Smith JE, Memenis SK, Holekamp KE (2007) Rank-related partnerchoice in the fission-fusion society of the spotted hyena (Crocutacrocuta). Behav Ecol Sociobiol 61:753–765

Stevens JR, Gilby IC (2004) A conceptual framework for nonkin foodsharing: timing and currency of benefits. Anim Behav 67:603–614

Tiddi B, Aureli F, Schino G (2012) Grooming up the hierarchy: theexchange of grooming and rank-related benefits in a new worldprimate. PLoS One 7:e36641

Vahed K (1998) The function of nuptial feeding in insects: a review ofempirical studies. Biol Rev 73:43–78

Vanak AT, Gompper ME (2009a) Dietary niche separation between sym-patric free-ranging domestic dogs and Indian foxes in central India. JMammal 90:1058–1065

Behav Ecol Sociobiol (2017) 71: 107 Page 13 of 14 107

Vanak AT, Gompper ME (2009b) Dogs Canis familiaris as carnivores:their role and function in intraguild competition. Mammal Rev 39:265–283

van Noordwijk MA, van Schaik CP (2009) Inter sexual food transferamong orangutans : do females test males for coercive tendency?Behav Ecol Sociobiol 63:883–890

Ventura R, Majolo B, Koyama NF, Hardie S, Schino G (2006)Reciprocation and interchange in wild Japanese macaques:grooming, cofeeding, and agonistic support. Am J Primatol 68:1138–1149

von Bayern A, de Kort S, Clayton N, Emery N (2007) The roleof food- and object-sharing in the development of socialbonds in juvenile jackdaws (Corvus monedula). Behaviour144:711–733

Wittig RM, Crockford C, Deschner T, Langergraber KE, Ziegler TE,Zuberbühler K (2014) Food sharing is linked to urinary oxytocinlevels and bonding in related and unrelated wild chimpanzees. ProcR Soc B 281:20133096

Yamamoto S (2015) Non-reciprocal but peaceful fruit sharing in wildbonobos in Wamba. Behaviour 152:335–357

107 Page 14 of 14 Behav Ecol Sociobiol (2017) 71: 107